Moving to a farm or giving up on the industry for this is basically saying the tools are hurting more than helping.
Tool fatigue is real. Most companies and developers do not invest in their SDLC. So you end up with a system where even a small change can require outsized effort leading to burnout. Tool fatigue is real.
Many libraries/languages/frameworks encourage a simple, vanilla SDLC. Think `rails` or `poetry` or `make`. But rarely does that scale once a project gets more than two developers, more than one repo, or more than one way of declaring dependencies. Everything turns into a snowflake script running a build-system integrated with hundreds of yaml files spreading out over repos and toolchains. Good luck doing fast end-to-end iteration on a change to something three layers deep in your dependency stack.
I wish the industry would try to solve this in a way that doesn't have vendor, language, or cloud lock-in.
Maybe some combination of containers, nix, asdf, and bazel or something. It doesn't seem that hard to be honest, just requires some innovation and resources to find and execute on an extensible and sane SDLC toolkit.
Until then: Find a way to iterate on your core problem. Become a jedi in at least your smaller domain. Focus on developer productivity. Make it easy to make a change. If it's not, find out why, and solve that problem exactly once rather than having to solve it for every change. Easier said than done. Often times it's politics and momentum that you'll have to overcome. Hopefully your company values this work.